基于约束多目标骨干粒子群的污水处理过程优化控制

Optimal control for wastewater treatment process based on constrained bare-bones multi-objective particle swarm optimization algorithm

为了取得污水处理过程能耗和出水水质这对冲突目标之间的最佳平衡,提出一种基于约束多目标骨干粒子群的污水处理过程智能优化控制方法。首先,利用数据驱动的思想建立能耗和出水水质的模糊神经网络预测模型;其次,利用带自适应扰动的约束多目标骨干粒子群优化算法对溶解氧和硝态氮浓度的设定值进行动态寻优,并利用模糊隶属函数法设计智能决策系统用于从Pareto最优解集中确定最优设定值;最后,利用模糊逻辑控制器实现底层跟踪控制。基于国际基准平台BSM1实验结果表明,建立的数据驱动模型能够有效辨识污水处理过程;同时,所提的多目标优化控制方法在保证出水水质参数达标前提下,能够有效地降低污水处理过程的能耗。

To achieve an optimal balance between energy consumption（ EC） and effluent quality（ EQ） of a wastewater treatment process（ WWTP）,an intelligent optimal control strategy is proposed based on constrained bare-bones multi-objective particle swarm optimization algorithm（ CBBMOPSO）. First,a data-driven-based fuzzy neural network prediction model of EC and EQ is constructed utilizing the process date measured from WWTP. Then,the proposed CBBMOPSO with adaptive disturbance is used for dynamically optimizing the set-points of dissolved oxygen S_O concentration and nitrate nitrogen S_（NO）level. Furthermore,an intelligent decision-making system based on fuzzy membership function is designed to identify the optimal setting value from the Pareto optimal set. Finally,the optimization setpoints of S_O and S_（NO） are tracked by a fuzzy logic controller to realize multi-objective optimal control of the WWTP. The experimental results based on the international benchmark simulation model No. 1（ BSM1） demonstrate that the proposed CBBMOPSO method can significantly reduce the energy consumption on the premise of assuring water quality.